 Okay, we are back live at Hadoop Summit. This is the end of day two. We're getting down to the home stretch here, Silicon Angles, continuous coverage of the big data world, the center of all the action, Silicon Valley, in San Jose, California, where we were having the first Hortonworks event of Hadoop Summit last year is run by Yahoo, where our guest was an employee, actually the day before last year's event. I'm John Furrier with Jeff Kelly here, Silicon Valley Wookiee Bonds Cube, Eric Baldish Wheeler, okay, got it. You were the CEO of Hortonworks as part of the spin-out and one of the co-founders, we've had almost all the co-founders on except for Shuresh, doesn't not come on. We're going to try to get him later, but we wanted to get all the founders on, in particular you, because you were involved actively, you were the acting CEO during that and publicly we're pretty vocal like, hey, you know, I'm a long-term CEO, I'm a tech guy. Remember your speech on stage last year, very humble. Lots changed since that year. So, 80 people now, growing like crazy. Growing fast. What's it like? Tell us what your year was like. Well, I mean, it's just, I think I've learned more every week than I learned in a year anywhere else. We've gone from 24 employees to 80. We've built out the product, opened up a couple new open source projects and Bari was something that had been started, but, you know, we'd just, we'd never really taken it to production anywhere. So, we've added a lot of new capabilities and now we're taking that out as part of HDP. We've built out all these connections with talent and putting that in our products. So, opened up a lot of different doors. We've built out sales, marketing, training, support, consulting. It's just, it's wild year. So, a lot of people don't know, a lot of people might not know. Some people do know, but might not know that Wharton Works was a big part of Yahoo's. The core team was a big part of Yahoo's Hadoop operation. Yahoo still has a bunch of people over there doing amazing work, but all that expertise wasn't like, it was like three guys in a garage or eight guys just kind of like, hey, let's get in the Hadoop business. You guys have been there, done that. And kind of had a shadow over you guys because Yahoo kind of wasn't stepping up and taking a lot of public credit for the Hadoop movement outside of the industry. But you guys did a lot of the work over there. I wouldn't say there was a shadow over us. I would say that if anybody who was in the Hadoop community knew what was going on, Yahoo took it from a prototype to what it is today. We put about 300 person years into Hadoop development over the first six years of the project. So I think that cast a pretty big invisible shadow. We've been running the summit for five years. So yeah, but we certainly felt that a year ago, it had reached a point where there was a market for Hadoop way beyond Yahoo and that there was an opportunity for us to take what we've been doing at Yahoo, take it to a much wider market and frankly, build a business that was self-sustaining that would then bring value to a wider community and to Yahoo. As the new employees come into Hortonworks, what's the culture like? I mean, describe the culture at Hortonworks. I mean, Rob was on theCUBE, was great. You can see it in his eyes, he's competitive. People have a fire in their belly at Hortonworks. Is that the culture? I mean, and how do you get the new employees in docinating that culture? We have our manifesto, which is on the website. But I mean, really stand out, I would say, first and foremost, relentlessly optimistic. We've, open source is an interesting game. There's nobody, when people leave your team, they continue to make your product better. Your competitors are working feverishly to make your product better as well. So we really believe that a key to this thing is being optimistic, looking for win-wins and looking to understand how we can do things with everyone in the community. And we're also very customer-centric, right? And that's very important, right? Ultimately, if you don't have customers, you don't have a business. So we're very focused on outcomes, we're very focused on delivery. You know, we're very focused on teamwork, right? Ultimately, we want people who understand, again, that if you're trying to build communities, you got to make space for other people to succeed. So there's a lot of that. Let's see, other things, you know, I've been in a lot of places that don't leave time for family or life outside of work. You know, if you don't have your health, if you don't have your family, you're not, you know, this is a marathon, not a sprint. So we try to make time for people outside of work. We're not a cult. We think that people should have a life outside of work. Right? Which, you know, finding that balance and a startup could be challenging because there's more to get done than, you know, trying to do it. I mean, you guys are like, it is marathon. Open Source is ultimately a marathon where everyone's watching. It's all out in the open. Even more with social media and connectedness. Have you guys looked, seen any changes in the collaboration side of the business? I mean, Open Source has always been collaborative. We had used NetSuite, had all kinds of online. But now everyone's online, everyone's connected. Has that changed the speed and some of the dynamics? Oh, I mean, tremendously. I mean, we're working real time with people all over the world. You know, it is, I mean, the whole Open Source movement, I think, has just been facilitated by that, right? I mean, we've got very simple central infrastructure that's just moving this thing forward, you know? And we're learning, you know, I mean, the thing, it's interesting. You know, I mean, the thing with Open Source is almost the failures are as interesting as the successes, right? And so you can put out this piece of software. A thousand people will try a thousand different things. And some of them will fork it and take a different directions. You'll see a few successes and those are great. You can pull those back. But also you can fail much more quickly, right? A lot of you see, you know, there's this group exploration of all the things that don't work. And so you learn a lot faster because you can try things in parallel in a way that you can't do it if it's a closed product. The good news is failure is accepted and people understand, especially developers, understand that, you know, when you write code in particular or try even startups, that is part of the learning. It is, well, I mean, it's, I should be careful Eric said that open-source software is all about failure. I don't want to leave that message. But when, if you think about our... That's not, I don't think we don't want to put that message out. It's not true. It's about iteration. It's about community. And yeah, and if you think about the whole Jeffrey Moore talk about crossing the chasm, right? There really are these two different constituencies. And there's a lot of people who don't want nothing other than to pull their bits off Apache, add their own special secret sauce and just try the new thing every week, right? And so, Apache supports that. We work with those guys every day. Some of our biggest partners are people that just have their own dev teams and we're learning from each other. Then on the other hand, you know, there's people that want to consume something tested and stable and that's what distributions like our HDP are for. Right? Let's take into that a little bit. So you've got a product out. I think we should ship tomorrow. So, you know, talk a little bit about, well, first of all, you know, why did you guys feel the need to really create your own distribution? And specifically, I love to talk about Ambari and really the role that plays because of course, you know, monitoring managing your deployment is critical and one of the areas that a lot of early adopters were struggling with. So maybe we can talk about first, kind of the why you felt it was necessary and then let's dig into Ambari a little bit. Sure, why the Hortonworks data platform? Well, I mean, the key thing for us is to provide, as I said, a good, simple experience for our customers where they get a dependable product that works. And for that, it's all about integration which means you want to test thing A with thing B with thing C. So you assemble a stack and you test it just the same way that you see, you know, Red Hat doing that or any of the other Linux distributions or, you know, so it's just, it's really about testability, repeatability, stability and deployability. That's why you have a distribution. And given that we're very interested in moving the projects forward and, you know, we needed to have a vehicle that where we could, you know, package, you know, our learnings and put it out to the market. So that's why we run the distribution business. In terms of Mbari, I mean, I think it is key, right? If there's one thing, all of the sort of, you know, if you look at Yahoo, we had a sort of tool set that we'd built up over a decade for managing clustered applications. And when Hadoop was built up, we used that proprietary toolkit for everything from installing to monitoring. And that stuff isn't in the market. So a lot of people were struggling, right? So, I mean, a lot of what we've heard, you know, talking to customers is, well, I like what Hadoop can do, but what is it doing, right? And how do I start? So, you know, we just tried to take the best practices that we'd seen and wrap those up into a consumable package. So that's what Mbari is. So on the deployment side, we're using open source, you know, we're using the puppet language and scripting system. And that's something that a lot of IT shops are very comfortable with. When we talk to people, they like that approach. And then on the monitoring side, people with Hadoop experience, it's sort of a lot of them had come down and said what had sort of picked Nagios and Ganglia, which are again two very popular open source projects that handle alerts and then real-time monitoring. So what we did was just try to take these best practices and wrap them up into a consumable form. And that's what Mbari is. And that's sort of where we're starting the conversation. And then we want to get that out there, solve some problems and we're going to keep it ready. We were talking with Sean and Sean Connolly, the strategy guy and he's talking about the team. We want to add some more Clydesdales as his direct quote. He's got the good expressions bringing doorbells, Clydesdales. He's got the good rap for the cube. I want to ask you about your, as you expand your team, because you've been that far from beginning, what is Ari Zilka's role as the chief product officer? You have the VP of engineering, you got a chief strategy. You got a bunch of guys who've been working together before. Has he been part of the team? He's been part of the conversation pretty much since our inception. He has been. Yeah, we were talking to him about the possibility of joining us for quite a while and we were lucky the stars aligned and that worked out. Now he's somebody that I've known for a number of years as have a number of other people on our team. And so his role is pre-sales and product management. So he helps us define what's in a release. He works very closely with our customers and he's really, really good at understanding a customer's set of problems and tools and identifying how our technologies can solve that. He's got an incredible breadth of experience, especially in the financials and retail sector. And he worked at Walmart for quite a while and his last company where he was CEO and then CTO was Terracotta, which is heavily used in the financial sector, right? And also in retail for sort of data acceleration. So he's going to be in charge of the roadmap too? Yeah, he's very deeply involved in the roadmap, very deeply involved in sort of figuring out what we can do, you know, what the customer needs. He's got a great intuition because he's used all these things. Okay, got it, cool. He's going to come on later. We want to make sure we drill down on him. I just followed him on Twitter. In terms of, so you know, the open source world, in terms of what can we expect in terms of product releases and upgrade cycles and how do you kind of view that? And you know, is that a yearly thing? Is that a more iterative approach? How do you approach that? Well, I mean, so one of the values that distribution vendor can bring is some amount of predictability. So the HTTP one platform will iterate every quarter. And you know, that is something that we can, you know, that everyone can count on. And it will take, you know, all the bug fixes for the problems that we've encountered in the field and innovations that have been tested and we can bring into that platform without disruption. So that's the kind of, you know, predictable train. Right? And then, you know, the other side of it is, you know, the sort of bleeding edge of all these projects can be very wild west. As I said, there's a lot of experimentation and some failure happening there. 2.0 is an alpha and that means it's being tried in a very few sort of bleeding edge places. And so it's a little hard to predict when that will close. We are, you know, as I said, using it in some places where people really value the technology, but it's hard to predict when it'll kind of converge and meet all of the acceptance criteria of the people we're doing those tests with. But, you know, guesstimation, I would say, that it should be, I'm expecting it to sort of go GA early next year at a guess. It might be earlier, right? But, you know. Well, yeah, I mean, we're looking ahead. I mean, we're, you just went GA and we don't want to look past that. I mean, that's a big milestone now. It is. And, you know, I mean, again, it's a stack. GA of a component is different from when does it make sense to sort of say, if you value stability, now is the time that it's worth it. I think what we're going to look to do is take it into very specific use cases where people value the new features. I mean, the two things that 2.0 really brings are A scale, right? When we started the project, I mean, this project has, we've been working at it for three years inside Yahoo and inside Hortonworks and people from around the community have contributed as well. But, scale was a real focus. You know, we want it to be able to go from 4,000 node clusters of the machines that were available three years ago to 10,000 node clusters and the machines that are available this year and next. So that's more like an order of magnitude. If you think about it, because the machine, the Moore's law is working in the background. So big ambitious jump in terms of how much storage you can manage, how much compute you can do in a cluster. So there aren't too many people who are frustrated that they can only run 4,000 nodes. But the people who want to go higher are very serious about that, right? Cause that'll have big operational return for people like Yahoo and Facebook and a select list of others. Then the other short term feature is just some improvements in performance. But frankly, 1.0 is pretty good and incrementally getting better as well. So then the rest of it is sort of what does this enable? It enables new frameworks. And I think that's where the real excitement is going to be in a year. MapReduce.do.o and those kind of things. Yarn specifically and I mean Federation as well. But Yarn will let us, everyone on the show wants to know how am I going to combine Hadoop with real time analytics, right? And Yarn does not in and of itself solve that, but it provides a framework in which we can start to do those experimentation, that experimentation we were talking about earlier. It'll let all the different actors who have a good idea plug their new compute frameworks into the same Hadoop clusters. And it'll let you try a lot of different things without changing the kernel, right? So, because we're moving sort of the programming model up to a place where users can plug that in per job. Eric, we got to get, we're getting the hook here. Eric Wallace for the CTO of Hortonworks Go Founder. Final comment that I'd like you to leave the audience with is obviously you guys are growing and very successful, peddling as fast as you can, getting to the point where you're right at number two with Cloud Air, quote in the peer, like all peer play, I guess, distribution, doing great work. What are you looking for in the team as you grow and expand, share with the folks key open positions, key things to fill, product, product map. Go ahead. Well, so as CTO, Sean's looking for Clyde Stales because we got a lot of work to do. I'm looking for Dragon Slayers, right? We've got to think about what comes next and really, we've got lots of really interesting problems to solve as we take these, as we solve the 2.0. I mean, there's all these different dimensions. So I'm looking for the sort of tech leaders who can work with us to solve this next big set of problems. You guys are doing great work. Obviously you're not slacking, you got a big investment. We hear more money's coming. We hear you guys are well-funded. Obviously the business deal's going on. Congratulations. What a year Hortonworks CTO co-founder, Eric. But thanks for joining theCUBE. We'll be right back with our next guest after this short break.